… but Blender is a 3D app first and foremost. So I doubt that there will ever be a real attempt to fix this 2D problem.
Hmm, I don’t think we can make that distinction in Blender anymore, it has and is moving towards the complete pipeline tool. Yes, 3D creation is the primary “mode”, but as you know the Open Movies are pushing technical targets, this time gearing Blender for a VFX production. If there isn’t an attempt to fix this 2D problem when it indeed needs a 2D solution, trying to do VFX in Blender (in the quality everyone expects) will result in loss of hair and smashed keyboards.
There’s only additional stuff needed to be built around this.
I still don’t think this code will work as a base for future roto-tools. Using 3D methodology and workflow for a 2D problem is trying to fit a square peg in a round hole.
I still have faith in Mango and that we will eventually get a proper roto tool and node. That will probably take 6-12 months, so if you desperately need a feathered mask in Blender now then sure, this code and a more involved process can perhaps work. All I’m saying is, let’s not take this code and say “job done”.